Deep Learning Profiler 21.11 Release Notes

Description

DLProf Release for 21.11, available in the NVIDIA TensorFlow 1.x, TensorFlow 2.x, and PyTorch NGC containers, and as a Python Wheel on the NVIDIA PY Index.

DLProf v1.8, which will be included in the 21.12 DLFW containers, will be the final release of DLProf. Starting with the 22.01 containers, DLProf will no longer be included. It can still be manually installed via a pip wheel on the NVIDIA PY index.

Driver Requirements

Release 21.11 is based on NVIDIA CUDA 11.5.0, which requires NVIDIA Driver release 495 or later. However, if you are running on Tesla (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), or 460.27 (or later R460). The CUDA driver's compatibility package only supports particular drivers. For a complete list of supported drivers, see the CUDA Application Compatibility topic. For more information, see CUDA Compatibility and Upgrades.

New Features

The key features of DLProf v1.7.0 / r21.11 are:
  • Released in the TensorFlow 1.x 21.11, TensorFlow 2.x 21.11 and PyTorch 21.11 NGC containers.
  • Added support for new Group Op view in the DLProf Viewer.
  • Removed support for custom NVTX domains.

Known Issues

  • This software is only supported for TensorFlow 1.15., TensorFlow 2.6, PyTorch 1.10, and TensorRT 8.0.3.
  • Collecting I/O data can cause issues in some cases, so it is off by default in this release. To turn it on, pass the following option to DLProf CLI:
    --nsys_opts=“-t cuda,nvtx,osrt -s cpu”

Resolved Issues

  • None.